Skip to content

Latest commit

 

History

History
21 lines (13 loc) · 937 Bytes

File metadata and controls

21 lines (13 loc) · 937 Bytes

Suggested Improvements to "Engineer" the Project

1. Security & Sandboxing (Highest Priority)

Currently, the tool executes AI-generated code directly on the host machine. This is a major security risk.

Feature: Implement execution inside a Docker container or a WebAssembly (WASM) runtime.

2. Virtual Environment Isolation

The project currently uses the system pip to install dependencies.

Feature: Automate the creation of a temporary Python venv for every script generated. This prevents the bot from "polluting" the user's global Python installation.

3. Local LLM Support (Ollama/Llama.cpp)

The project currently relies on HuggingFace's API.

Feature: Add a "Local Mode" using a tool like Ollama or a Rust crate like candle (by HuggingFace).

4. Static Analysis (Linting)

The project currently uses py_compile to check syntax.

Feature: Integrate ruff or flake8 to perform deeper static analysis.